MRI Compatible Incubators

Introduction

For over two decades at University College London there has been use of an in vivo model to investigate neonatal hypoxic ischaemic encephalopathy (HIE) and assessing possible new therapies to protect the brain. This study is unique in that it uses Magnetic Resonance Imaging (MRI) and Magnetic Resonance Spectroscopy (MRS) both for outcome biomarkers but also as a real-time measure of experiment parameters. In each experiment the subject is scanned on three consecutive days, quantitatively tracking the effectiveness of a trialled therapy whilst receiving intensive care.

The combination of ancillary equipment with an MRI scanner can be challenging as there are many factors to consider. If the equipment is made from materials with magnetic properties (for example steel) then this will be attracted into the magnet bore, and present a very dangerous projectile hazard. The force on magnetic components can also prevent them from performing their function correctly, for example a steel spring in a valve. The MRI scanner creates intense time-varying magnetic fields at both radio frequencies and audio frequencies, which the equipment will need to be protected from. It is also a very sensitive radio receiver, and interference generated by equipment will result in image artefacts.

Intensive care requires continuous physiological monitoring, anaesthesia and ventilation, and temperature control. In the absence of MRI compatible versions of the equipment, it is common to keep the intensive care equipment outside of the magnet room, passing all non-conductive tubing through the waveguides. However this requires disconnecting this tubing first, which adds unnecessary risk to the subject during transport.

The development of the MRI compatible incubators (in collaboration with University College London and Chiesi Farmaceutici S.p.A.) was motivated by the need to simplify the transport procedure to and from the MRI scanner, and also to allow the use of multiple MRI scanners: a pre-clinical 9.4T MRI scanner, and a 3T clinical MRI scanner. The pre-clinical system is favoured for its high field strength which is more optimised for in vivo imaging, however given the translational nature of the study, using a clinical MRI scanner has obvious benefits as the same sequences can be used that will be used in clinical scanning of babies for HIE. This necessitates a mobile solution, incorporating all of the life-support systems required for intensive care, all of which must operate within the vicinity of an MRI scanner.

Our initial requirements specification called for simplification, automation, mobility, and MRI compatibility. Due to the limited availability of suitable MRI compatible equipment and the need to control all internal systems from a simple and intuitive user interface, we took the opportunity to custom develop much of the sub-systems, allowing for a high degree of specialisation and optimisation. These custom parts include:

  • An aluminium framed trolley with integrated radio-frequency interference (RFI) shielded enclosure. The enclosure incorporates waveguide and bulkhead penetration points for tubing and cables, and a copper finger lined door, allowing for easy access to the internal systems without compromising the shielding integrity when closed.
  • A specially designed subject holding pod that has optimal loading into both the 9.4T and 3T MRI scanners.
  • A novel anaesthesia and ventilation system that enables the use of two different ventilators; an on-board MRI compatible transport ventilator, and an externally provided intensive care ventilator. Pneumatically operated valves seamlessly switch between these two ventilators.
  • A wireless, MRI compatible physiological monitoring system was designed and built.
  • Each incubator contains a linux powered single board computer that handles all communication between devices, and serves a web graphical user interface that can be accessed on any HTML5 compliant browser/device (i.e. Windows, Mac, Linux, iOS, Android) to display all relevant information about the subject's physiology, and the status of the incubator’s systems.
  • A temperature control system for the subject, comprising of a heated water circulation system, and control software that adjusts the water temperature in response to the subject's temperature.
  • A power switching system that enables most of the incubator’s systems to be powered either by 12V DC derived from on-board mains power supplies, or from a large 100Ah 12V lead acid battery. When mains power is disconnected the battery automatically kicks in resulting in no power loss to any of the on-board systems.

Incubator Systems

The incubator holds all of the life support, environmental control, monitoring and auxiliary systems required to deliver intensive care at all times. This has been split into the following systems:

  • Server and Communications System.
  • Physiological Monitoring System.
  • Environmental Control System.
  • Ventilator System.
  • Power System.

In addition the shielded box inside the incubators provides room for up to 8 standard clinical infusion pumps, for delivering drugs and other infusions such as saline. By integrating these pumps into the incubator, it removes the need to ever disconnect them. The pumps were located at the furthest end of the incubator from the MRI scanner, ensuring that they experience minimal static magnetic field, and so the operation of the motors within each pump is not affected. The infusion lines run through waveguides located at the front of the magnet

Server and Communication System


The incubator’s sub systems consists of a set of distributed, self-contained modules that each perform a specific function, for example a heated water circulated pump, physiological monitoring module, a power switching module etc. Each of these contains a microcontroller, and is fully self-sufficient for operation. To enable central control and collation of data (such as measurements, or status updates), each device has a built in serial interface which communicates with a central server. This server consists of a small single board computer, running a linux distribution. A built-in resistive touchscreen display, mounted behind an optically transparent RFI filter provides a continuous interface to the server’s GUI, even during MRI scanning.

Serial data is sent via plastic optical fibre (POF), rather than using conductive cables. This has two benefits. First, it removes the chance for high frequency noise to be couples onto signal cables, for example from clock sources, reducing EMI within the incubator’s shielded enclosure. It is far easier to shield an electromagnetically quiet environment, and also means that in the even the shielding fails (for example through mechanical damage, or a door gets inadvertently left open), the effect on the MRI images will be less severe. Secondly, it also eliminates the chance for ground loops to occur between connected equipment, reducing the potential for data corruption.

Implementation of the POF network was achieved by including a POF transmitter and receiver on each custom module’s PCB, and also by developing a POF-TTL module, for use in integrating off-the-shelf equipment equipped with a serial interface, and also for interfacing to USB-Serial adapters at the server.

Physiological Monitoring System

The physiological monitoring system measures the subject's blood pressure, ECG, temperature and airway pressure, and then sends this data via Bluetooth to the incubator’s server. It has been built with non-magnetic materials, and incorporates design features that prevent the intense EMI generated by the MRI scanner from damaging the internal circuitry, and also prevent any noise generated by the internal circuitry (for example from fast digital transitions) from being detected by MRI scanner’s sensitive receiver.

Standard disposable medical sensors for an invasive blood pressure transducer, thermistor temperature probe, and ECG electrodes connect to a RF filtered breakout connector. This mates with the data acquisition module, using robust LEMO connectors via a short length of shielded multicore cable. Any radio frequency interference picked up by the sensors will be shunted via the filters to the shield of the multicore cable. Once the signals are on the PCB they pass through some additional filtering, transient voltage suppression diodes for removing large, potentially damaging voltage spikes, and then pass into a copper-can shielded region via some feed-thru capacitors.

Two air pressure transducers are also included, with non-magnetic stainless steel connectors on the outer enclosure. They can measure pressures up to 4kPa, making them ideal for monitoring the tracheal pressure, and the movement of a respiratory motion pad.

Each sensor’s signal path then has an analogue front end circuit for amplification and conditioning, before going to a TI ADS7870 12 bit ADC. Data is sampled at 500Hz for each channel, and is then sent over SPI bus to the microcontroller, an Atmel ATMEGA 328P running at 14.7456MHz – a clock frequency that enables the use of high speed UART rates with minimal error.

The microcontroller handles setting up the ADC for each conversion, and also generates packets of data to be sent via Bluetooth over a serial port profile. Our chosen Bluetooth module connects via a high pass filter (2.3GHz corner frequency) to an antenna mounted on the outside of the data acquisition module’s aluminium enclosure.

In addition to the data acquisition module, pulse oximetry is provided by a Nonin 7500FO MRI compatible pulse oximeter. This has a standard serial interface, and by using a POF-TTL adapter this was connected to the server so that the data can be logged and displayed.

Finally, the data acquisition module is powered by a 1300mAh lithium-polymer battery pack. Under normal operating conditions this provides in excess of 20 hours of battery life, more than enough for a long scanning session.

Environmental Control System

Temperature regulation is provided by a heated water circulation pump. The requirements for this were that the water temperature can be set by the central server, allowing for both manual and automatic regulation of the subject's core temperature by the server. MRI compatibility, and portability meant that using mains power for the pump and heating element was not possible.

An off-the-shelf thermal therapy pump was modified to meet our specifications. This allowed us to re-use the water-tight enclosure and brass heat exchange block, significantly saving design and manufacture time.

  • The mains operated, and highly magnetic pump/motor was removed and replaced with a submersible 12V DC powered pump.
  • A 40W 12V cartridge heater with ¼ NPT connector was custom made to fit into the supplied brass heat exchange block.
  • A 60W 12V submersible, flexible rubber heating mat was put into the water tank to provide additional heating power.
  • A custom PCB was designed incorporating a microcontroller, high current mosfets for control of the heating elements and pump, thermistor analogue front end circuitry, and POF communications.

While still containing magnetic materials, the submersible was tested to work in the vicinity of the MRI scanner, and is small enough to be held securely down. Using 12V components means that the pump can operate even while the incubator is moving.

The microcontroller runs a PID (proportional, integral, derivative) control algorithm that maintains the water temperature to within ±0.1°C of the target temperature.

Ventilator system

Continuous ventilation and inhalation anaesthesia is a critical part of intensive care, and so a major requirement of the incubators was that this should always be available. While MRI compatible ventilators are available, these are only suitable for transport, and are not as fully featured as an intensive care ventilator. Therefore it was important to develop a mechanism two switch between an on-board MRI compatible ventilator and an externally located intensive care ventilator. We achieved this by using pneumatically actuated 3/2-way valves, one for the “to subject” flow of fresh gas and one for the “from subject” flow of exhaust gas. A manual valve easily accessible at the rear of the incubator provides the actuation pressure, derived the on-board high pressure supply of medical air. Using pneumatically operated valves, as opposed to solenoid valves ensures full operation in a magnetic environment.

Non-magnetic gas cylinders provide a portable supply of oxygen and medical air, with the option to plug into an external source of high pressure gases if the incubator is parked for long periods of time. A full anaesthesia system is provided, incorporating an isoflurane vaporiser, fraction of inspired oxygen (FiO2) meter, and an airway humidifier. FiO2 measurements are sent to the server, and are logged by the physiological monitoring software. We also designed and 3D printed some custom connectors for connecting between pneumatic tubing and standard medical ventilation hoses

Power System

The incubator’s sub systems were designed to run from a 12V bus, which is provided by a mains powered adapter when the incubator is parked, and a 12V lead acid battery when the incubator is mobile. The power system consists of the battery, 12V power supply, battery charger, and a custom designed power switching controller that automatically switches between 12V sources as required. This controller also reports the power usage by the incubator, enabling an estimation of the battery life. Each sub system has its own circuit breaker, isolating electrical faults so that they do not cause a complete system shut-down.

Software development

The software and system infrastructure was designed from scratch with the main goal of being as accessible and easy to use and possible. The two primary pieces of functionality that were desired from this was the detailed physiological monitoring of the subject, and the control of the life support systems.

An initial design decision that was made was to develop the system architecture around cutting edge web-based technologies. This allows any compatibility with multiple types of devices which can access the monitoring and life support systems concurrently, and more importantly, allow any mobile device with a modern browser to also do so. This allowed us to supply the research team with a set of android tablets, with which they can use to check the subjects’ physiological data, perform message logging, be aware of any alarm signals, control the temperature control of the subject, and review any data over the current or any of the previous experiments. The web- based approach also allows remote access to this functionality, allowing senior members of the team to access the same data from a remote location with any standard browser.

The server is designed to accumulate and visualise all of the important vital signs in real time, both in a waveform visualisation, and numerically, so that the research team may understand the current state of the subject and the system with a quick glance at their tablet. Algorithms have been designed and implemented to allow real-time processing of ECG, blood pressure and respiratory signal into heart and respiratory rates. A hardware signal gating system allows interfacing to the MRI scanner in order to perform cardiac or respiratory gating for better MR image quality.

Configurable alarms allow audio alerts whenever the vital signs are raised above or lowered below user defined thresholds. This allows the team to respond quickly to events such as a spike in blood pressure, or a decreasing respiratory rate.

One of the core requirements of the project was to remove as much manual interaction with the system in order to automate as many tasks as were possible. This would allow the team to focus on the research rather than menial tasks. An example of a task which the team previously had to carry out was the manual adjustment of the temperature of the water bath in order to keep the subject at a desired constant temperature. This involves the monitoring of subject temperature levels and frequent adjustment of water bath temperature. This labour intensive process is no longer required as the incubator software makes automatic adjustments to the water bath temperature, in order to regulate the subject temperature.

All data is logged at the minimum rate of once per second. The graphical user interface allows access to this data in the form a series of graphs which may be stacked and viewed over a user defined period of the experiment. This allows the user to easily assess the experiment at a particular event and compare what occurred with respect to the various physiological signals at that time. The user may also review any previous experiment and access the data both in graph visualisations, as well as in a spreadsheet format where the data rate is specified by the user and interpolated by the system before file creation.

A challenge in creating the system was to create a network topology and end-to-end communication protocol that allowed the individual components of the system to exchange messages and information with one another. The individual hardware components (UART) perform serial communication. This is passed through a serial to fibre-optic convertor to reduce the possible RF interference this may create. The central server communicates with the monitoring system over a Bluetooth interface which operates at a frequency range (2.4GHz) compatible with the MRI system, and allows the subject preparation and transportation to be fully wireless. The central server communicates with each of the clients over a Wi-Fi interface, again to be fully wireless and operational in a compatible frequency range. The central server serves a webpage through which the clients may connect, and allows communication over the WebSockets protocol allowing full duplex communication over TCP/IP. The protocol developed allows any device (from tablets down to the individual hardware components such as the temperature controller) to perform end-to-end communication with any other device, for such task as sending and acknowledging commands and streaming data. Concepts such as timeout and retransmission ensure that critical information, such as an adjustment of the subject temperature, is guaranteed to be received by the intended device.

Incubator Manufacture


The incubator has three main physical components. The first is a pod to house and support the subject. The pod is constructed from PVC and designed to fit within the preclinical MRI scanner bore and to align the subject with the most sensitive point within the scanner. The pod is also removable so it can be placed on the bed of a clinical MRI scanner. The second component is a truck to which the pod attaches, the truck allows the pod to slide forwards into the preclinical MRI scanner. The truck runs on rails attached to the third component; the trolley. The trolley is the bulk of the incubator system and houses the support machinery within a protective faraday cage (which prevents the scanner effecting and being affected by the equipment inside). The trolley also houses the computer system to control the equipment and to monitor the subject via various sensors. The trolley is mobile so that it can move between the human and preclinical MRI scanners.

The Pod


The pod is designed to allow the front section to be removed from the support section. This modular design allows the subject to be prepared in the front pod on a surgical table allowing greater ease of access. The front pod with the subject in place is then loaded into the support section of the pod. A rail on the bottom of the front pod aligns the front pod towards its locked in position. The support section of the pod includes two large handles so that the entire device can be easily moved about.

The truck

The truck supports the pod and helps with tubing management when the pod slides forward into the preclinical MRI scanner. The truck has a large central hole allowing the front pod to be loaded into the support pod while it is attached. The truck also has a simple heavy-duty latch to connect it to the pod. This latch can be opened and the entire pod lifted away from the truck to be placed on the bed of a human MRI scanner.

The Trolley


The trolley consists of a large aluminium welded frame housing a plate aluminium box with electromagnetically shielded bulkheads and doors to allow access. The trolley has space at the front for MRI safe gas canisters for the subject. MRI compatible caster wheels allow the entire incubator to be moved. An electromagnetically shielded touch screen allows users to directly interface with the on-board computer. The rear of the trolley has mounting points for ventilation support equipment for the subject. The internal space of the trolley houses:

  • Water heater and pump – subject temperature regulation
  • Air heater and blower – subject temperature regulation
  • Power control systems – battery and mains power control
  • Oxygen and anaesthesia switching apparatus – subject support gases
  • Humidifier – subject support gases
  • Large battery – mobile power source
  • Syringe pumps (x8) – drug delivery

All of the devices in the trolley were custom manufactured or converted to be MRI compatible.

Manufacture

The frame of the trolley was designed by us and outsourced to a specialist aluminium welder for fabrication. The pod and truck were designed and manufactured entirely in house.

Some specialist equipment was required to allow us to manufacture these custom pieces of hardware. Imgenious has a CNC router capable of cutting precise shapes from various materials including aluminium and PVC.

We made extensive use of 3D printing to produce both work jigs and structural components for the incubators.

Other standard workshop tools that we used on this project include; a lathe, a large circular mitre saw, a multi-axis pillar drill, a belt bench mounted sander and a myriad of the usual hand tools.